Skip to content
Just Say

Privacy

Your voice stays on your Mac.

Just Say is local-first dictation software. This page describes what the Mac app does. It does not add claims the app does not make.

Audio

  • The microphone is used while you hold the shortcut, use hands-free mode, or explicitly start Meeting Mode.
  • On Mac, transcription uses a local Parakeet model or Apple’s on-device Speech engine. German, Hindi, and Russian use a local multilingual model.
  • Temporary dictation recordings are deleted after transcription unless you turn on “Keep audio files.” Meeting recordings stay until you delete the meeting.
  • Just Say does not require an account, an API key, or cloud storage.

What can use the network

Internet is used for an optional download or repair of local speech models. That transfer is the model file. It is not your transcript. Installer updates are downloaded from this site. Once a day the app also asks GitHub whether a newer version is out. Nothing about you, your voice, or your text is sent, and you can turn it off in Settings → About.

Just Say does not send dictation audio or transcripts to OpenAI, Anthropic, Gemini, AssemblyAI, Deepgram, Whisper API, or any required cloud inference service.

Writing on your computer

English cleanup — spacing, grammar rules, Apple spelling, an optional on-device grammar model, style, and personalization — runs on the Mac. The optional grammar model is used only if you already placed its files locally. Just Say does not download it and does not send text out for grammar.

Dictionary entries, snippets, custom styles, and app profiles stay on your computer. Learning suggestions stay local until you confirm them. Passwords, cards, keys, tokens, and cookies are not stored. Export is a JSON file you choose. It is not sent automatically.

Where data lives

~/Library/Application Support/ShabdaTara/

  • Settings
  • History and notes
  • Meetings
  • Dictionary entries and custom styles
  • Snippets and app profiles, kept locally
  • Optional audio, if you turn on Keep audio files
  • Screenshots from a region you explicitly capture
  • Speech models cached on disk

Screen and meetings

Screen Recording is optional. If you enable it and circle an area while dictating, Just Say stores a local screenshot. Meeting Mode can capture system audio for that meeting only, and only after you press Start. Those files are not uploaded.

Meeting summaries are local and extractive.

Claims Just Say does not make

  • Just Say is not a cloud assistant and not a replacement for a large language model.
  • It does not claim to understand everything on your screen.
  • Mac installation uses JustSay-macOS.dmg from this site.
  • Windows installation uses JustSay-Setup.exe from this site.